{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:A32YYNXQ5GVVL5MMPMISENKD2M","short_pith_number":"pith:A32YYNXQ","schema_version":"1.0","canonical_sha256":"06f58c36f0e9ab55f58c7b11223543d33f4270677cd25ca0ad8b71281627028f","source":{"kind":"arxiv","id":"2605.15336","version":1},"attestation_state":"computed","paper":{"title":"HoloMotion-1 Technical Report","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"HoloMotion-1 trains a humanoid tracker on a hybrid mix of noisy video motions and clean MoCap data to achieve zero-shot whole-body control.","cross_cats":["cs.AI"],"primary_cat":"cs.RO","authors_text":"Bo Zhang, Kaihui Wang, Maiyue Chen, Qijun Huang, Xihan Ma, Yi Ren, Yucheng Wang, Zhiyuan Yang, Zhizhong Su, Zihao Zhu","submitted_at":"2026-05-14T18:59:43Z","abstract_excerpt":"In this report, we present HoloMotion-1, a humanoid motion foundation model for zero-shot whole-body motion tracking. A key innovation of HoloMotion-1 is to scale control-policy training with a large-scale hybrid motion corpus, where video-reconstructed motions from in-the-wild videos provide the dominant source of motion diversity, while curated motion-capture and in-house motion data provide higher-fidelity supervision and deployment-oriented coverage. This data regime enables HoloMotion-1 to move beyond conventional MoCap-only training and exposes the policy to substantially broader behavio"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.15336","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-14T18:59:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9aafcc7666f8a6dd055e4d571a7c7725a173eb455400008646056372542a7f6a","abstract_canon_sha256":"66e4cb458850217c84ad771a3f5c231c3477c311152a6ec9d3979f0b98a03d12"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:53.198703Z","signature_b64":"JbAFr+LezKbPfjh1RGNI9tloXRmmxqQSBNLenuUI1FypVRMWAUn3UOsZyJpwZgZFXHPVBKHg1dRZ9031lAiECA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"06f58c36f0e9ab55f58c7b11223543d33f4270677cd25ca0ad8b71281627028f","last_reissued_at":"2026-05-20T00:00:53.198043Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:53.198043Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HoloMotion-1 Technical Report","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"HoloMotion-1 trains a humanoid tracker on a hybrid mix of noisy video motions and clean MoCap data to achieve zero-shot whole-body control.","cross_cats":["cs.AI"],"primary_cat":"cs.RO","authors_text":"Bo Zhang, Kaihui Wang, Maiyue Chen, Qijun Huang, Xihan Ma, Yi Ren, Yucheng Wang, Zhiyuan Yang, Zhizhong Su, Zihao Zhu","submitted_at":"2026-05-14T18:59:43Z","abstract_excerpt":"In this report, we present HoloMotion-1, a humanoid motion foundation model for zero-shot whole-body motion tracking. A key innovation of HoloMotion-1 is to scale control-policy training with a large-scale hybrid motion corpus, where video-reconstructed motions from in-the-wild videos provide the dominant source of motion diversity, while curated motion-capture and in-house motion data provide higher-fidelity supervision and deployment-oriented coverage. This data regime enables HoloMotion-1 to move beyond conventional MoCap-only training and exposes the policy to substantially broader behavio"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments on multiple unseen motion benchmarks show that HoloMotion-1 generalizes robustly across diverse motion types and capture conditions, significantly improves tracking accuracy over prior methods, and transfers directly to a real humanoid robot without task-specific fine-tuning.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that video-reconstructed motions from in-the-wild videos can serve as the dominant source of motion diversity while reconstruction noise, source-domain mismatch, and uneven quality do not prevent effective learning from the higher-fidelity MoCap and in-house data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"HoloMotion-1 trains a large Mixture-of-Experts Transformer policy on a hybrid corpus of video-reconstructed and MoCap motions to achieve robust zero-shot whole-body tracking that transfers directly to real humanoid robots.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"HoloMotion-1 trains a humanoid tracker on a hybrid mix of noisy video motions and clean MoCap data to achieve zero-shot whole-body control.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f568e99a85ace95b395162ede44efcc7d0b1cc4b12397bece371724caecc1658"},"source":{"id":"2605.15336","kind":"arxiv","version":1},"verdict":{"id":"ba12836a-c980-4a60-a749-4a8d9d6135a2","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T15:59:21.032919Z","strongest_claim":"Extensive experiments on multiple unseen motion benchmarks show that HoloMotion-1 generalizes robustly across diverse motion types and capture conditions, significantly improves tracking accuracy over prior methods, and transfers directly to a real humanoid robot without task-specific fine-tuning.","one_line_summary":"HoloMotion-1 trains a large Mixture-of-Experts Transformer policy on a hybrid corpus of video-reconstructed and MoCap motions to achieve robust zero-shot whole-body tracking that transfers directly to real humanoid robots.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that video-reconstructed motions from in-the-wild videos can serve as the dominant source of motion diversity while reconstruction noise, source-domain mismatch, and uneven quality do not prevent effective learning from the higher-fidelity MoCap and in-house data.","pith_extraction_headline":"HoloMotion-1 trains a humanoid tracker on a hybrid mix of noisy video motions and clean MoCap data to achieve zero-shot whole-body control."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15336/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T16:31:18.284252Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T16:04:48.313689Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T14:41:54.188208Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.759080Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"48dcb5a117e7787de402628e4e47d9e06010ed0414d7393c25e1e80a40430399"},"references":{"count":27,"sample":[{"doi":"","year":2025,"title":"Gmt: General motion tracking for humanoid whole-body control","work_id":"6bbe3751-0317-41b9-9025-267f828466ed","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Guoqing Ma, Siheng Wang, Zeyu Zhang, Shan Yu, and Hao Tang","work_id":"c34be14c-d3bf-4313-a163-0dc964026ab6","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Track any motions under any disturbances","work_id":"798c1b8c-56b2-4a7b-8a7b-d82939b75862","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Kungfubot2: Learning versatile motion skills for humanoid whole-body control","work_id":"d7ed46a3-6119-46f0-ac17-3a2034df85a5","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"Language models are unsupervised multitask learners","work_id":"5fa609b3-1203-4f0e-a526-0110cb3f8046","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":27,"snapshot_sha256":"46fbb4e3fd793ecb7148cb1978c6d5cd8e0ac156f920f4b214e3933469deee8f","internal_anchors":2},"formal_canon":{"evidence_count":2,"snapshot_sha256":"2d5cd6b0ccb73272d991127aea22c4e13b53a8e90c1593bad4eba6d5448ae5f0"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.15336","created_at":"2026-05-20T00:00:53.198173+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15336v1","created_at":"2026-05-20T00:00:53.198173+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15336","created_at":"2026-05-20T00:00:53.198173+00:00"},{"alias_kind":"pith_short_12","alias_value":"A32YYNXQ5GVV","created_at":"2026-05-20T00:00:53.198173+00:00"},{"alias_kind":"pith_short_16","alias_value":"A32YYNXQ5GVVL5MM","created_at":"2026-05-20T00:00:53.198173+00:00"},{"alias_kind":"pith_short_8","alias_value":"A32YYNXQ","created_at":"2026-05-20T00:00:53.198173+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M","json":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M.json","graph_json":"https://pith.science/api/pith-number/A32YYNXQ5GVVL5MMPMISENKD2M/graph.json","events_json":"https://pith.science/api/pith-number/A32YYNXQ5GVVL5MMPMISENKD2M/events.json","paper":"https://pith.science/paper/A32YYNXQ"},"agent_actions":{"view_html":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M","download_json":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M.json","view_paper":"https://pith.science/paper/A32YYNXQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15336&json=true","fetch_graph":"https://pith.science/api/pith-number/A32YYNXQ5GVVL5MMPMISENKD2M/graph.json","fetch_events":"https://pith.science/api/pith-number/A32YYNXQ5GVVL5MMPMISENKD2M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M/action/storage_attestation","attest_author":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M/action/author_attestation","sign_citation":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M/action/citation_signature","submit_replication":"https://pith.science/pith/A32YYNXQ5GVVL5MMPMISENKD2M/action/replication_record"}},"created_at":"2026-05-20T00:00:53.198173+00:00","updated_at":"2026-05-20T00:00:53.198173+00:00"}